Classic Book Review: Philosophical Investigations


Philosophical Investigations
Ludwig Wittgenstein
Blackwell 1953, 3rd Edition 1967

ISBN 0631146709 (paper)

Buy it from Amazon.com (commission paid)
Buy it from Amazon.co.uk (commission paid)

Other Reviews


Tractatus Logico-Philosophicus
Ludwig Wittgenstein
Routledge, 2nd edition, 2001

ISBN: 0415254086 (paper)

Buy it from Amazon.com
Buy it from Amazon.co.uk
Einkaufen von Amazon.de

This little book is a surprise in many ways. Not least, it is short and easy to read -- at least, if you dip into it and read a page, you will have no difficulty seeing what Wittgenstein is talking about. The visual metaphor is intentional -- Wittgenstein spends a lot of time discussing what we mean by seeing and understanding; what we can talk about publicly, and what is private and forever inside our heads. In fact, another surprise is that the book is illustrated, in one place even in colour! If you don't know what Wittgenstein's duck-rabbit is, perhaps you now will -- showing is more powerful than telling (and there's a lesson in there for everyone involved in eliciting requirements). What does the figure mean, what does it show? That's another matter: reflect on it, or read the book (page 194ff).

And perhaps the book's title is less forbidding than the famous Tractatus Logico-Philosophicus -- though he writes in the Preface: 'Four years ago I had occasion to re-read my first book (the Tractatus) and to explain its ideas to someone. It suddenly seemed to me that I should publish those old thoughts and the new ones together ...' In other words, this book is a commentary and reflection on all of his thinking. It was written in the final years of his life, and published posthumously. The Tractatus is a numbered list of short statements or aphorisms, which makes it quite hard to read in the ordinary sense -- I don't know about you, but if I read a saying or a proverb, I have to stop and think, do I agree with this or not? What would it mean for me? Is it true? What implications does it have? and without this discussion written into the text, after a few aphorisms I have started so many hares I don't know where I'm going. But the Investigations presents ideas, gives examples, and discusses them, which is a lot more approachable; and since many of the ideas are from the Tractatus, this makes it a nice way in to that book as well.

It's a surprise too, that someone with a long german name should be so, well, English; so simple and direct; so clear; so interesting -- and so relevant to requirements. Normally, of course, one has only to mention something like Weltanschauung and everyone knows to switch off at once, their internal spam-filter cutting in to preserve an acceptable signal/noise ratio. So I will try to say in a few words why Wittgenstein should not be one of the long W-words among your 'trash' filters.

Writing requirements is about translating what people need but can't necessarily express, into a set of statements that developers can understand well enough to build something -- a system -- whose operation satisfies those people's needs. This task is hard not because we aren't clever enough at inventing gadgets and software (if anything we're too clever), but because the translation of needs into written meanings is far more than the clerical task (the word is Dijkstra's) of writing specifications.

Polanyi's maxim is 'We know more than we can tell', and this forms a convenient stepping-stone into Wittgenstein's thought, which ranges over questions of meaning and how we understand; of knowledge and consciousness; of language and logic; of sensory observation and communication; and more.

For example, how do we ensure that people on a project mean the same thing when we use a term like 'system reliability'? Obviously it matters, for instance if the contract that contains the requirements talks about measuring this quantity and rewarding the supplier accordingly. Wittgenstein says that

43. 'For a large class of cases .. in which we employ the word "meaning" it can be defined thus: the meaning of a word is its use in the language.

And the meaning of a name is sometimes explained by pointing to its bearer.'

Therefore, we had better make sure that the language in use in our project is shared by all involved in it -- for instance, we should make a dictionary, and we had better ensure that it corresponds closely to the language people already use; what is more, we must root out any other uses to avoid confusion. That might take a bit of doing? Well, yes; but the alternative is ambiguity and arguments over contracts. What was all that about the bearer of a name, then? We need to associate dictionary terms with things in the world -- by what Jackson calls 'designations' -- to form bridges that jump the gap between our formalised language and models, and the world that our stakeholders understand. Then if someone does not understand our requirements, we can point to the things we are talking about, and explain our meaning in terms of them.

Pointing perhaps doesn't sound very much like the subject of a Philosophical Investigation? But it is.

45. ... a name is not used with, but only explained by means of, the gesture of pointing.

Pointing, or ostension to give it its more general name, is a vital bridge between the world and what we and other people have inside their heads. How do we share meanings with other people? How do we know what other people see when there's a red letter-box in front of them? What if they're colour-blind, anyway? (The problem of 'qualia', or what we'd call attributes -- like redness -- of things in the world, is a deep one, and it doesn't go away.)

II xi. I look at an animal and am asked "What do you see?" I answer: "A rabbit". -- I see a landscape; suddenly a rabbit runs past. I exclaim "A rabbit!"

Both things, both the report and the exclamation, are expressions of perception and of visual experience.

Well, what if I wasn't familiar with rabbits: what would I have said then? What would I have seen? (Wittgenstein may also be hinting here at the writings of W.V.O.Quine, who also discussed how people could share the name of an animal by looking and pointing at and naming it, and what sorts of misunderstandings and ambiguities could arise -- but usually don't.)

Someone suddenly sees an appearance which he does not recognize. ... Is it correct to say he has a different visual experience from someone who knew the object at once?

As a birdwatcher, I can glance at a pond and see Pochard, Coot, Moorhen, Mallard, ... while a companion sees nothing special -- if asked, he might say 'Ducks' (partly right, and partly wrong: swimming waterbirds would be more accurate). So, does someone with more knowledge and more specialised language see things differently? What happens if we try to gather requirements when we are talking a) to people less expert than ourselves; b) to people more expert in the domain (banking, aerospace, whatever) than ourselves? Either way, there is going to be a difference in meaning, if not indeed also of perception.

Perhaps these few hints may have helped to give some idea of what Wittgenstein was thinking about, and how fresh and immediate he is, and how important to our discipline. I think I'll give him the last word. He begins his Philosophical Investigations with a quotation (in Latin, but I'll give the English) which beautifully summarizes the subject:

1. When my elders named some object, and accordingly moved towards something, I saw this and I grasped that the thing was called by the sound they uttered when they meant to point it out [ostendere]. ... Thus, as I heard words repeatedly used in their proper places in various sentences, I gradually learnt to understand what objects they signified...

{Saint} Augustine {354-430 A.D.}, Confessions, I. 8 {397 A.D.}

These words, it seems to me, give us a particular picture of the essence of human language. It is this: the individual words in language name objects -- sentences are combinations of such names. In this picture of language we find the roots of the following idea: Every word has a meaning. This meaning is correlated with the word. It is the object for which the word stands.

© Ian Alexander 2003